Challenging the "End of Growth" Argument

Our latest podcast, “Why America’s Economic Growth May Be (Shh!) Over,” is based on a muchdiscussed paper (abstract, PDF) by the Northwestern economist Robert J. Gordon. It is a fascinating paper that is well worth a read, and its provocative argument has certainly gleaned a lot of attention.

Roger Pielke Jr., whose field of expertise is science and technology policy, has read the paper and argues that Gordon is (what’s the best way to put this?) wrong:

Over the past month I have taken a close look at Gordon’s paper, the data he relies on and the papers that he cites.  My conclusions are that Gordon’s analysis is deeply flawed and tells us essentially nothing about the potential for future economic growth. It does help to reveal a big gap in the discipline of economics, and that is the utter lack of an explicit theory of growth and the mechanisms by which it actually takes place. What Gordon has provided, in his own words, is a “a provocative fantasy” one that tells us much about the discipline of economics but little about the state of the world.

I am not much for making predictions, but I wouldn’t be shocked if Gordon’s paper soon inspires a public forum or two on the topic and that Pielke is invited to rebut.


It seems that growth simply must taper off. There are surely limits to what most people can want, and a lot of those wants are not great contributors to the economy. If I can fulfill all my economic wants on $X per year, at some point beyond that I stop contributing to growth.


Needs are finite; wants are limitless. Even the people with millions of dollars of net worth are still trying to keep up with their more-affluent Joneses. And at the other end of the spectrum, you've got billions of people in developing countries who are looking at the middle-class Western lifestyle saying, "Why can't I have that too?" Wants and desires are gonna continue for a LONG time. Probably as long as there are people.

Resources, on the other hand, ARE finite. Yes, we can squeeze more out of what we've got through efficiencies or finding new raw materials, but I can guarantee that the limiting factor to growth is going to be resource scarcity, not desire satisfaction.


I have to disagree. For me, and I think for a lot of people, wants are finite. Or at least there's a big gap between living comfortably and being able to start your own space program, in which there are few if any wants to be filled, and where filling them is often more trouble than pleasure, For example, I could easily buy a new car or computer every year or two, but why should I go to the trouble, when the ones I have (and am used to) work just fine?

You can see this at work even now, in for instance the slowing sales of new cars & personal computers, and in the growth of advertising which tries to persuade people that they really need to buy this or that new thing.


As long as there are barriers to free exchange of goods and ideas around the world, removal of these always can add some growth potential for everyone.

Also, we have a long way to go toward understanding all the physical constraints of our universe. As our scientific knowledge expands, there will be staggeringly more variations on uses for our resources to satisfy wants. Very likely paradigm shifts will occur that we cannot imagine today.

Eric M. Jones.

I agree with Roger (but haven't spent the $5 for the paper). Thinking that the universe has run out of steam is a fool's game, but explaining how to measure growth is key to a reasonable definition of progress.

If you measure wealth by the number of cars each of us has in our driveways, or the amount of food we eat, the gadgets produced or even the GDP/capita, you would be limiting the definition of progress. Fuller thought that progress could be measured in kW-Hrs/capita, but this was before the computer revolution. Perhaps now we can measure progress as Gigabytes/second/person.

The coming of Ray Kurzweil's "singularity" might redefine progress overnight. Conscious computers, Superintelligences, virtual reality indistinguishable from reality; hey I'm for it, whatever it is.

I often tell the story of my friend's Grandmother who as a young girl in 1884, sailed from Australia to LA in a four-masted wooden sailing ship. Not a single paved street existed in LA then and cowboys and Indians were its main inhabitants. She lived to see astronauts walk on the Moon and thought it wonderful. My friend though that we could not see any similar advances in our lifetimes.

He was wrong.


Mark Bahner

"The coming of Ray Kurzweil’s “singularity” might redefine progress overnight."

I think that is exactly what's going to happen. In fact, as I just noted, I think the world per-capita GDP is going to begin increasing by more than 10 percent per year within the next 10-20 years, due to advances in artificial intelligence.

That may seem totally insane, but it's not at all. I used Ray Kurzweil's estimates of computing power per $1000, and multiplied them by the number of $1000 personal computers expected to be produced every year, and came up with this graph:

It's a graph of the number of "human brain equivalents" (HBEs) added every year by personal computers. The graph explains not only why computers have not added significantly to economic growth to this point, but why they *will* add significantly to economic growth in the very near future.

Here are the number of human brain equivalents (HBEs) I calculated were added to this point:

Only 1 HBE in 1996.

Only 1000 in 2006.

I predict 1 million HBEs will be added in 2016. But that's still fair from significant, because we have approximately 75 million actual human brains that are added every year, due to human population increase. But where the curve (and resulting economic growth) really kicks off is:

1 billion HBEs added in 2024, and
1 *trillion* HBEs added in 2033.

The economist Julian Simon noted that (free) human minds are what cause economic growth. And Ray Kurzweil has predicted that computers will soon match and then greatly exceed the capabilities of the human mind. Therefore, it's quite reasonable to predict that world per-capita economic growth will really take off (greater than 10 percent per year) in the next 1-2 decades. Virtually no economist (with the notable exceptions of Arnold Kling and Robin Hanson) has picked up on this likelihood, even though it will be the most important thing that has ever happened in economics, and even though it likely will happen in the next 1-2 decades.

P.S. My prediction that world per-capita GDP will be increasing by more than 10 percent per year within the next 1-2 decades if off if Terminator-like artificial intelligence develops, or if there is a global thermonuclear war.


Mark Bahner


Roger Pielke Jr. isn't the only non-economist ripping into the economics profession. ;-)

From those comments:

So I think that, somewhere between approximately 2020 and 2030, world per-capita GDP will begin increasing at over 10 percent per year. If it does, Arnold Kling should definitely get a Nobel Prize in Economics. (But he probably won't. Nobel politics! ;-))

It is astonishing to me how little the economics profession appears to understand the profound impact on economic growth that’s likely to occur as the result of inexpensive human-level artificial intelligence.

From those comments:

Someone needs to point the economics profession to Julian Simon's observation that (free) human minds are what create wealth, and to Ray Kurzweil's observation that very shortly computers will approach (and then vastly exceed) human minds


Mark Bahner

"If I can fulfill all my economic wants on $X per year, at some point beyond that I stop contributing to growth."

Some people are like that. But look at Mark Zuckerberg or Warren Buffet. They certainly don't need the money, but they go to work every day. Heck, Warren Buffet is 82 years old.


Sure they go to work, just as I do: not because I really need the money or desire the stuff I could buy with it, but because I enjoy what I do. In fact, from what I've read of these two men's lives, they could be poster children for the idea that many people can satisfy their wants on middle-class incomes.

Beyond that, there are wants that are, or can be, satisfied at a reasonably modest, and fixed, cost. For instance, I like to spend time hiking, biking, riding the horse, etc. This does cost a bit of money, but it's going to stay pretty much the same over time. Likewise, I like to listen to music, but there is only so much I can listen to: 16 hours worth per day (assuming I don't listen while I sleep). Trying to listen to two or more pieces simultaneously just doesn't work, so there's a basic upper limit to growth there.

Then there's the matter of physical limits to economic growth, which are closer than many people might think, as explained here:


Mark Bahner

Hi James,

You write, "Sure they go to work, just as I do: not because I really need the money or desire the stuff I could buy with it, but because I enjoy what I do. In fact, from what I’ve read of these two men’s lives, they could be poster children for the idea that many people can satisfy their wants on middle-class incomes."

Yes, I agree completely. But the conclusion I get from it is exactly the opposite from yours. I think the fact that people work even though they don't need the money means that the economy will grow long beyond when people need money just to get by.

For example, the world per-capita GDP is currently a little over $10,000. I'm predicting that the world per-capita GDP will be over $10 million in the year 2100. But what I think that really means is something like Star Trek where people just do whatever they want, and money doesn't really come into the matter. What I don't think is properly depicted in Star Trek is that I think robots will do any jobs considered unpleasant or dangerous (e.g. garbage collection or mining).

"Then there’s the matter of physical limits to economic growth, which are closer than many people might think, as explained here:"

Tom Murphy (the finite physicist) is completely wrong, as explained here and here:

In fact, I suspect that Tom Murphy knows there's no limit to world per-capita GDP, because he never says what he thinks the limit is.

If you know Tom Murphy, let him know that I'd be happy to debate him on his blog. We could call it: "Exponential Engineer meets Finite Physicist."


Mark Bahner

"Take trash collecting: my local company runs trucks with a crew of 3. Say they’re paid $50K/year each, or $1.5 million in 10 years. Now replace them with a $500K robotic trash collector that’ll last 10 years before wearing out. Hasn’t the size of the economy just been reduced by $1 million?"

Good question. A minor comment/correction...the $1 million is over 10 years, or $100,000 a year.

I would maintain that the "total *value* of goods and services" (the GDP) in the economy has not declined...unless the garbagemen (oops, garbagepersons) become alcoholics, rob a bank, and get put in prison.

Let's say that the garbagepersons go from 50k per year to 20k per year jobs. So their total salary has gone from 150k per year to 60k per year. But I maintain that the total *value* of the goods and services has increased, because the machine is giving the exact same *value* of service as the garbagepersons, and the garbagepersons are now doing 20k per year jobs that they would otherwise not have done (of course, if they're displacing other 20k per year people, that needs to be taken into account). Where I think the accounting really breaks down is that the decline in the cost of garbage collection needs to get accounted for in the Consumer Price Index.

I think we agree is that it's all very complicated. For instance, here are just some of the possibilities I see as a result of computer-driven cars, buses, and trucks:

1) Car ownership becomes unnecessary. There's no need to have your car sitting in a garage or parking lot, if you can order *exactly* the right type of car when you need it and where you need it. Most people drive their cars less than 2 hours out of every 24 hours. So the number of cars and trucks purchased each year could easily be cut by a factor of 5. And most people make compromises in what they drive (e.g., buying a minivan to haul kids, but also using it for single-person driving). No need for such compromises with computer-driven cars. Order a minivan when you need one, and a single-person car when you need that.

2) No need for parking lots. No parking lots at the mall, grocery store, business, or stadium, because everyone is delivered door-to-door by computer-driven cars. No parking lots means a lot less stormwater run-off.

3) No need for lights at the parking lots. No need for most streetlights, in fact. Light pollution is tremendously reduced.

4) No need for garages at home or parking garages in cities. All home garages can be converted to playrooms, workout rooms, whatever.

4) Speeds are dramatically increased. Cars, buses, and trucks can be driven bumper-to-bumper at 100+ mph on freeways without accidents.

5) No need for car insurance. No car accidents.

6) Stores become essentially obsolete. Your groceries or clothes or whatever can be delivered to your door. Delivery charges are minimal, because the delivery vehicle can carry the produce for many customers on a single run. No stores means no need for Walmart and Target supercenters, with all their frozen food aisles.

7) Old people, young people, and disabled people don't need to worry about mobility.

8) Streets can be dramatically narrowed. In fact, they can be two strips of asphalt with grass in the center and trees/shrubs right up close to the road. Or people can walk on the edge of the road without fear of being hit by a careless driver.

9) Short-haul airplanes become less attractive. If the speed on the interstates is over 100 mph, without any driving hassle, it's less attractive to take a plane on a trip of a few hundred miles. High-speed rail also becomes less attractive. (Note that the 100+ mph interstate trip need not be by single-occupant car, but could instead be by a mini-bus or full-sized bus.)



"But I maintain that the total *value* of the goods and services has increased, because the machine is giving the exact same *value* of service as the garbagepersons, and the garbagepersons are now doing 20k per year jobs..."

Possibly, but I think it's far more likely that most of those $20K/year jobs have also been filled by robots, so the former trash collection crew are on the dole. Indeed, isn't that pretty much what we've seen happen with lots of manufacturing jobs - with the caveat that it's been cheaper to hire Chinese than to buy robots. (But now the Chinese are buying the robots...)


One could equally argue that the invention of farming makes all subsequent inventions minor in comparison. If you had the choice to return to hunter gathering or give up electricity, which would you choose?


I'd take being a hunter-gatherer. So much easier when you've got a fridge and freezer.


While amusing in its own way, the paper is fatally flawed because it is looking from the middle of the information revolution. We are still very far from understanding the impacts of fully integrating information technologies within our lives. That revolution is not the Internet bubble and it is hardly over. Visit the MIT Media lab website and get a glimpse into what might be coming in 10 years or so. These are practical application of existing technologies - no breakthroughs or 'singularity' needed. Moving further into the future, we could all become cyborgs before it's 'over'!

Beyond that in a more speculative but still very plausible realm, the biological - nanoscale manufacturing technologies will give us new and smart materials. New materials always presage new products and often new industries ( imagine a flexible and high efficiency photovoltaic materials - it could cover every surface available... ).



The question you need to ask about all that gee-whiz stuff from MIT's Media Lab and elsewhere is what fraction of the population is actually going to want most of it.

Then you need to ask whether having any particular gadget actually increases the size of the economy. As for example my cell phone: a $7/month gadget replaces a $30/month land-line (and also reduces the chance that I'll purchase another digital camera, voice recorder, etc). So hasn't that reduced the economy by $28/month?

Likewise your cheap photovoltaic material: if I buy some, I give the economy a one-time shot of whatever the purchase price is, but reduce the ongoing economy forever by the amount of my monthly power bill.

Mark Bahner

"While amusing in its own way, the paper is fatally flawed because it is looking from the middle of the information revolution."

Yes, the chessboard story is very applicable to this situation. In the chessboard story, a wise (though maybe not socially wise) man asked for payment from a king of one grain of rice (some witnesses say wheat) on the first square of a chessboard, two on the second square, 4 on the third square, 8 on the fourth square, and so on. The total amount of material on the chessboard ends up being about 1000 times the total world production of rice in 2010.

Ray Kurzweil made the point that total worldwide Internet connections have been doubling every few years for decades. But it didn't matter when the went from 100 to 200, or 200 to 400, or 400 to 800, and so on. But when they go from 100 million to 200 million, people start to notice. And when they go from 1 billion to 2 billion, they're huge.

A similar, but even more powerful, story will unfold for artificial intelligence. According to my calculations, wer'e still only adding computers with the capacity of a few hundred thousand human brains every year. That's not noticeable when we're adding 75 million human brains by population increase. But by 2024, I calculate we'll be adding the equivalent of 1 billion human brains every year by computers. And by 2033, it will be 1 *trillion* human brains added every year.

So in the next 10-20 years, I predict we'll begin to experience what Ray Kurzweil calls "the second half of the chessboard" with regard to artificial intelligence and subsequent economic growth.